Chaos Library?
Hi,
I'm interested in knowing if someone has ported André Sier's Chaos Library (A-Chaos Lib) to Max 7 or intends to do.
Thanks!
Hi Daniel,
I worked on non-linear dynamic system a couple of years ago.
I remember there was some libraries about, but now I can't remember and didn't like them so much.
Making some "chaotic oscillators" (or audio rate non-linear dynamic model) in max/gen~ is not really complicate. You just have to pay attention interpretating differential systems ,)
I attach an example below - Lorenz Non-Linear Dynamic System.
I also made some externals (and they worked!), but I never finished them, I'm sorry..
hi - I made 64-bit versions of some of the a-choas lib externals for my own usage, but didn't finish the whole lot. It's not difficult to do, just needs some time. If there is interest in this, I could to the rest and make osx versions of the lib available - just don't know when that will be.
But most of the stuff in the library could also be done in max directly.
Please note, that there are no audio objects in a-chaos, it's all message based.
btw, did you ask the original author about updated versions?
Hi, Luca and Volker,
Thank you for your time and answers!
Your patches helped, Luca.
It really doesn't seem so difficult to port the library and I will give it a try. Would you mind sharing the a-chaos externals you have been able to port, Volker? I'm specially interested in the Hénon-Heiles attractor, which I've already started recoding from Sier's source code (from here http://s373.net/code/ > at the end of the page about a-objects). I've also written to him, as you suggested.
There are problems in implementing a strange attractor with objects in a patcher. I have tried a lot before. You can solve iteration at the cost of speed and precision, but more complicated equations, as Luca said, need a real work out, that is not a game for non-mathematicians.
Thanks again, and will post any new infos!
Daniel Puig
Would you mind sharing the a-chaos externals you have been able to port, Volker?
I'll see what I have over the weekend.
I'm specially interested in the Hénon-Heiles attractor, which I've already started recoding from Sier's source code
As far as I remember the implementation of henon-heiles is buggy in the a-chaos lib.
By the way, all objects of the original lib will work if you put max in 32bit mode.
There are problems in implementing a strange attractor with objects in a patcher. I have tried a lot before. You can solve iteration at the cost of speed and precision,
What do you mean exactly? Since max uses 64-bit floats, precision is not an issue anymore and I guess you wouldn't see a big difference in speed either - at least for message passing.
The thing to watch out for and which makes implementations of recursive stuff in max a little cumbersome, is order of execution.
Here is a implementation of a lorenz attractor in max:
https://cycling74.com/forums/zl-lorenz-pmpd-granola
it's rather old and max was still 32-bit, but nowadays, the output from a patcher solution vs an external should be pretty much the same in this case.
vb
I'll see what I have over the weekend.
Thanks!
As far as I remember the implementation of henon-heiles is buggy in the a-chaos lib. By the way, all objects of the original lib will work if you put max in 32bit mode.
Since max uses 64-bit floats, precision is not an issue anymore ...
True as for the buggy implementation and all other observations. I've used it a lot and knew the reset problem, so it was easy to track down the bugs in the code already. Thanks for pointing all that out and the link for the discussion. I've realized it, after checking out the other solutions. Since 2015, owning a 2010 MacBook Pro for performance, I have chosen to stay in Max6, and lost track of gen~. Restarting it now.
Hope to post a patch here soon with Hénon-Heiles implemented. All the infos here, examples, Sier's source code, have helped a lot!
here is a temporary link to a 64-bit built for macos of the a-chaos lib.
http://vboehm.net/xchange/A-chaos-Lib_64-bit.zip
no bugs fixed yet, just made all objects work in 64-bit mode.
might do a proper release somewhat later.
Many thanks !! very useful for me. cheers
Indeed!!
Many Thanks, Volker!!
Here is the Hénon-Heiles System in a gen~ patch and a maxpatch with UI that I've finished today. Sorry for not sending the gen~ patch before. It works well, but can surely be improved. Would you mind looking through the patches? It seems there is some code that could be ripped off.
I'm working on finishing an implementation with a generalized Hamiltonian. The version we know, and in the included patches, assumes a two-dimensional Hamiltonian.
The contact with André Sier has been very useful and he has been so kind to share some more code. Should we start a project here and work together on a library? It would be great to work with Volker, Francisco and all.
All the best and thanks for the sharing!
Daniel Puig
*** These are the equations used:
*** gen~ patch
*** maxpatch
It would appear that your main patch is missing some things....
Hi, Max and all,
Would very much appreciate any ideas :) I'm working on the UI.
Here is the generalized hamiltonian gen~ patch. If you call it on the other patch, initial conditions should work with the default C (-1.) and lambda (1.). This article states the integrable values for C and lambda. It really works well. The 5th output is now total energy of the system and I still haven't figured out how to get the potential. The additional outputs are polar coordinates inside gen~ (radius and angle > angle has to multiplied by 100) that I'm experimenting with.
This article by Vernov has helped figuring out the mathematics.
http://theory.sinp.msu.ru/~svernov/Papers/TMF2003.PDF
All the best!
Daniel
Posting again the main patch. Hope it works now!
It seems to work fine now, Daniel. I also took the liberty of replacing your uh... long bit with a compressed version of the patch (select what you want to copy, choose Copy Compressed from the Edit menu, and paste the result in).
Thanks, Max!
So, these are the last gen~patch (henonheilesGH.gendsp), maxpatch (henonheilesGH-02.maxpat). I've included the ambimonitor external (by ICST, https://www.zhdk.ch/5381 ) in the patch, to visualize polar trajectories of (x, y) and (xt, yt). The gen~patch gives radius and angle of the two pairs in the 4 last outputs. Sure there are other ways to do it, with jsui or jitter... but the gen~ works.
I really hope to implement other non-linear dynamics' models. Thought of looking at more up-to-date versions of strange attractors and fluid dynamics. Do you know of anyone working in that direction that would be interested in collaborating?
All the best,
Daniel
I wish I could contribute some gen~ patches, but due to my limited math skills, I have only converted the original a-chaos-lib help files and some needed patches to max 7. I made a simple overview patch, too. All should work together with volker’s 64-bit versions.
best
ben
Hi everyone!
Coming from a physics background where I've studied dynamical systems, I'm just getting into Max (8) --> anybody still around on this forum? I'd love some tips for getting started :) !!
Probably the most important thing to understand in doing dynamical systems in max is implementing iterative structures correctly.
For message-based iteration you need to be absolutely careful with message order, updating variables and triggering of calculations. I'd recommend using one master [metro] object which would synchronize all your calculations. Avoid using [send] and [receive] for transmitting synchronization bangs, because they break the message-ordering queue (ordering becomes indeterminate).
Max message order works right to left, top to bottom. There are objects you can use to force a different message order, like [trigger]. With multiple variables it can be useful to [pack] them into a list to help with syncing up variables (you can then [unpack] them later).
Also the [expr] object is valuable. You can send a synchronized list to its left inlet instead of individual variables to discrete inlets, to help ensure all your variables have been updated synchronously.
If you are in the signal domain, implementing feedback loops properly is vital. Most likely you'd be using gen~, so you can use single-sample delay, which has the [history] object for that.
If you are using matixes in jitter, you need to remember that matrixes are passed by reference, which can be very convenient to simplify otherwise patches that would otherwise potentially be messy and confusing.
Hope that helps.
EDIT: here's an example using [expr] without it triggering at the wrong time:
@UNAMI TAKU
i'm late but - thank you so much! i just did a quick test and it looks very good!
Hi, bit of a noob here so excuse me if I'm missing something obvious, but how do you get these into Max? I'm on 64bit windows max 8.
I put the a-chaos-lib_win folder in the library folder (tried packages also) but when I try to use the objects I get "Error 126 loading external <object name>"